Silicon is made by heating quartz rock with carbon in a massive electric furnace at 1,500 to 2,000°C, which strips the oxygen from silicon dioxide and leaves behind molten silicon. That single step produces what’s called metallurgical grade silicon, which is about 98–99% pure. Getting from there to the ultra-pure silicon inside your phone or solar panel requires several additional rounds of chemical purification, each one pushing purity higher by orders of magnitude.
Step 1: Extracting Raw Silicon From Quartz
The starting material is high-purity quartz rock, not ordinary beach sand. While beach sand is mostly the same mineral (silicon dioxide), it contains too much iron, aluminum, and other metal contamination. Manufacturers select quartzite with the lowest possible impurity levels to give the process a head start.
Inside a submerged arc furnace, this quartz is mixed with carbon sources like coal, charcoal, or petroleum coke. Massive electrodes pass electric current through the mixture, generating temperatures between 1,500 and 2,000°C. At those temperatures, the carbon pulls oxygen away from the silicon dioxide in a reaction called carbothermic reduction. The simplified version: silicon dioxide plus carbon yields liquid silicon plus carbon dioxide gas. The molten silicon collects at the bottom of the furnace and is tapped off into molds.
The result is metallurgical grade silicon, rated at 1 to 2N purity (meaning 99% to 99% pure). That’s good enough for making aluminum alloys and silicone chemicals, but nowhere near clean enough for electronics or solar cells.
Step 2: Purifying Silicon With Chemical Reactions
To go from 99% pure to 99.9999% pure or better, manufacturers convert the solid silicon into a gas, clean that gas, then convert it back into solid silicon. The dominant method is called the Siemens process.
First, crushed metallurgical grade silicon is loaded into a fluidized bed reactor and exposed to hydrogen chloride gas at around 350°C. The silicon reacts with the hydrogen chloride to form a compound called trichlorosilane, a liquid that boils at a low temperature. This is the key trick: because trichlorosilane evaporates easily, it can be distilled repeatedly, and each round of distillation strips away more impurities. Trace contaminants like boron and phosphorus form their own chloride compounds, which separate out during distillation because they boil at slightly different temperatures.
Once the trichlorosilane is sufficiently clean, it’s mixed with hydrogen gas and fed into a chemical vapor deposition (CVD) reactor at roughly 300°C. Inside this reactor, thin silicon “seed” rods hang vertically. The trichlorosilane decomposes on contact with these hot rods, depositing pure silicon layer by layer, like frost building up on a cold window. Over many hours, the rods grow into thick, high-purity polysilicon cylinders. The final product reaches 99.9999% purity or higher.
Purity Grades and What They’re Used For
Silicon purity is measured on the “N scale,” where each N represents a nine in the purity percentage. The three main tiers are:
- Metallurgical grade (1–2N): 99% to 99% pure. Used in steel and aluminum alloys, and as the feedstock for further purification.
- Solar grade (7–9N): 99.99999% to 99.9999999% pure. Clean enough for photovoltaic cells that convert sunlight into electricity.
- Electronic grade (9–11N): 99.9999999% to 99.999999999% pure. Required for computer chips and semiconductor devices, where even a few atoms of contamination per billion can ruin performance.
The jump from solar grade to electronic grade is enormous in terms of effort and cost, even though the numerical difference looks small. Each additional nine requires increasingly aggressive purification steps.
Turning Polysilicon Into Usable Wafers
Polysilicon rods or granules aren’t directly useful for chips or solar panels. They need to be melted and reformed into a single, continuous crystal structure. The most common method involves melting the polysilicon in a crucible, then slowly pulling a seed crystal upward from the surface of the melt. As it rises, silicon solidifies around it in a perfectly ordered atomic lattice, forming a cylindrical ingot that can weigh over 100 kilograms.
These ingots are then sliced into wafers using diamond-coated wire. The wafers for solar cells are typically just 0.15 to 0.2 mm thick, sliced from ingots with a 156 mm cross-section. At that thinness, breakage is a real concern. Wafers cut to 0.15 mm break at a rate of about 6%, while slightly thicker 0.2 mm wafers break only about 2% of the time. Each cut also wastes material as sawdust (called kerf loss), so manufacturers are constantly working to use thinner wire and reduce that waste.
Doping: Adding Controlled Impurities
Pure silicon is actually a poor conductor of electricity. To make it useful in electronics, manufacturers intentionally add tiny, precisely controlled amounts of other elements. This process is called doping.
Adding phosphorus atoms gives silicon extra electrons, creating what’s known as N-type silicon (the N stands for negative charge carriers). Adding boron creates holes where electrons are missing, producing P-type silicon. The boundary between an N-type region and a P-type region is what makes transistors and solar cells work.
Doping can happen in two ways. During crystal growth, small amounts of the dopant element are mixed into the molten silicon. For more precise control, especially in chip manufacturing, ions of phosphorus or boron are fired directly into the surface of a finished wafer using a machine called an ion implanter. This shoots the atoms at energies around 80,000 electron volts, embedding them to a depth of about 1.2 micrometers below the surface.
Byproducts and Environmental Costs
The Siemens process generates a significant byproduct: silicon tetrachloride. For every kilogram of pure silicon produced, several kilograms of this compound come along as waste. Silicon tetrachloride is highly volatile, acidic, and corrosive, making it a serious environmental hazard if released untreated.
The polysilicon industry has developed several strategies for dealing with it. The most common is hydrolyzing it (reacting it with water) to produce a material called white carbon black, which has uses in rubber and coatings but is a low-value product with limited market demand. Some facilities convert the silicon tetrachloride back into trichlorosilane and feed it back into the production loop, which reduces both waste and raw material costs. Researchers have also found ways to transform it into specialty catalysts and other higher-value chemicals, but these applications are still niche.
A Cheaper Alternative: Fluidized Bed Reactors
The Siemens process dominates the industry, but it’s energy-intensive and slow because it works in batches. An alternative approach uses a fluidized bed reactor, where tiny silicon seed particles are suspended in an upward stream of silane gas. The silane decomposes on the surface of the particles, coating them with pure silicon until they grow heavy enough to fall out of the gas stream. This produces small, spherical granules instead of thick rods.
The energy savings are dramatic: fluidized bed reactors use 80 to 90% less electricity than the Siemens process. They also run continuously rather than in batches, and their granular output is easier to handle and melt in downstream processing. Despite these advantages, the Siemens process has held its market position for decades because it produces higher-purity silicon and has benefited from widely available cheap electricity. As energy costs rise and solar panel demand grows, fluidized bed technology is increasingly seen as the leading candidate for producing lower-cost polysilicon at scale.